This generates test cases using formal techniques like boundary value analysis, equivalence partitioning, and pairwise testing. You feed it constraints (like age 18-120) or parameter combinations (browser/OS/screen) and it cranks out minimal test sets with maximum coverage. The qe-test-generator agent handles the grunt work of identifying boundaries and reducing redundant tests. Good when you're staring at complex business rules or too many parameter combinations and don't want to write hundreds of overlapping test cases by hand. It's basically automating what you learned in that software testing course but probably don't apply systematically because it's tedious.
npx skills add https://github.com/proffesor-for-testing/agentic-qe --skill test-design-techniques